Estimation of photometric redshift (photo-z) is more time- and resource-efficient than finding redshift through spectroscopic surveys. Photo-z estimation uses statistical techniques or machine learning to learn a mapping between the color and the redshift of the galaxies. Because photo-z estimation is trained using knowledge of galaxies with both spectroscopic and photometric measurements, if the galaxy sample is incomplete in color space then systematic error may be introduced. In this project we explore methods for visualizing and identifying incompleteness in galaxy redshift training datasets. The Self-Organizing Map (SOM) is a useful tool for reducing dimensionality and visualizing similarity in a dataset, and we explore its usefulness in characterizing the effect of the aforementioned incompleteness. SOMs provide a convenient way to visualize the higher dimensional color space of the photometric dataset. We investigate how reliably the SOM can detect incompleteness in both color and redshift space using statistical tests to quantify the photometric redshift error produced by the incompleteness.